You might try this method, developed by Chen et al. (2010):

http://onlinelibrary.wiley.com/doi/10.1002/chem.201000441/pdf

Chen, Y., Wu, Y., Henklein, P., Li, X., Hofmann, K. P., Nakanishi, K. and Ernst, Oliver P. (2010), A Photo-Cross-Linking Strategy to Map Sites of Protein–Protein Interactions. Chem. Eur. J., 16: 7389–7394. doi:10.1002/chem.201000441

To summarize, they started with p-benzoyl-phenylalanine (BPA). BPA is a photo-inducible crosslinker that resembles phenylalanine. You can engineer an ORF for your protein in which one or more Phe codons are replaced with amber stop codons. You can then express this ORF in an E. coli strain that expresses a special amber anticodon tRNA and aminoacyl tRNA synthetase that will charge this special tRNA with BPA. Importantly, you will have to supplement the E. coli growth media with BPA. Ordinary E. coli will grow very slowly since many housekeeping genes with amber stop codons will be improperly translated. However, I believe there now exist E. coli strains in which all amber codons have been switched to the other two stop codons. Then, the only limiting factor would be whether your protein aggregates upon BPA substitution for Phe.

If your protein expresses and purifies well, you can simply incubate it with its suspected binding partners, activate BPA with UV (350nm), separate complexes by SDS-PAGE, trypsinize, and perform tandem mass spectrometry on the peptides.

This is a fairly standard procedure that has been successfully implemented by multiple research groups, although optimization can be tricky. You mentioned that you would want to cleave complexes with DTT. What would be the purpose of this cleavage? I mentioned the Chen et al. (2010) paper above because they followed BPA modification with further derivatization to afford a click-chemistry ready adduct that could be modified with biotin and purified with strepavidin. This might work for you if you require immuno/affinity precipitation. However, you would probably want to use a peptide of your protein of interest rather than the whole protein since a protein would probably not be amenable to the organic solvents used for the multi-step synthesis.

I cant do this. Unfortunately I don't have so much time and its not possible to express my protein in E.coli and purify it. I want to identify binding partners in vitro in my cell lysate of a colorectal cancer cell line. I am using DTME because it does only modify cysteine. It should be cleavable because of mass spectrometry. So what I actually want to do is to cross-link possible binding partners of that protein in the colorectal cancer cell line, perform immunoprecipitation of this protein and analyze the eluate via mass spectrometry. To be sure that trypsin digestion is not influenced by the cross-linker I can only use cysteine-cross-linkers. It has to be cleavable to be able to identify those proteins by mass spectropetry. Otherwise some part of the binding partners would still stack together and this could lead to wrong identification of peptides. We have no idea about interactors of this protein. I tried Co-Immunoprecipitation without cross-linker. However we know that the binding affinity of this class of proteins is not so strong so we were afraid that I will loose binding partners during washing steps. That's why I decided to use a cross-linker. In one experiment I was able to cross-link my protein in vivo and precipitate afterwards it with magnetic beads (Flag-M2 coupled).I splittet the eluate by half and was running one half under non-reducing conditions (linker is intact) and the other half I was running under reducing conditions (linker is splitted) on sds-page. Under non reducing-conditions I observed a shift of my protein. However when I repeated this experiment I was not able to get the same results again. I saw that my protein is cross-linked but I was not able to precipitate it anymore with magnetic beads. I don't know what happened between those two experiments. Unfortunately there exist no recombinant version of my protein up to date.

Eva, you can use amine-reactive crosslinker and mass spec to find your partners. Crosslinking is never done to saturation and it would modify only a small number of peptides and it won't create any problems for protein ID. We all routinely use it in the lab. Amine-reactive crosslinkers are the best to capture potential interactive partners because Lys are quite abundant amino acid and often found on the protein surface. DSP is cleavable, in case you need it.

Mapping of exact crosslinked site is very challenging especially without knowing the interacting proteins. But based on what you said, you may not need it. 

I was also thinking about changing the cross-linker. In many articles I read that they used amine-reactive cross-linkers and mass-spec. So you think its also not a problem with Trypsin digestion prior to Mass-spec? I hope I can convince my supervisor..

Eva, you just need to test different crosslinker concentration to find the one where you can see a crosslinking product but not too much where "everything is crosslinked to everything."  If you don't over-crosslink, there will be no problems with trypsin digest Surely, there will be breaks in coverages with some peptide missing, but it rarely gets into the way of protein ID. Amine-reactive crosslinkers are more efficient as Lys are more abundant. In contrast, Cys are not that abundant, but also need to be in a reduced form for crosslinking. 

Ok thank you so much for your help.  I hope I can convince the mass spectrometry core facility  :-)

I agree with Anna K and we have used a water soluble, amine reactive, thiol cleavable crosslinker (DTBP) with MS. See our papers:

https://www.ncbi.nlm.nih.gov/pubmed/19738201

https://www.ncbi.nlm.nih.gov/pubmed/25727331

https://www.ncbi.nlm.nih.gov/pubmed/26833789

Best wishes

Jon

Thanks a lot. We will try it. I hope it will work :-)